Danang Architecture University

The increasing sophistication of cyber threats necessitates the use of advanced techniques like Deep Learning (DL) for Network Intrusion Detection Systems (NIDS. While DL models achieve high accuracy, their “black-box” nature hinders their adoption in real-world security operations, where transparency and trust are paramount. This paper investigates the practical challenges of applying Explainable AI (xAI) to this problem by critically analyzing the outputs of a standard xAI framework on a high-performance NIDS model. We use a Deep Neural Network (DNN) trained on the widelyused CIC-IDS-2017 dataset as a baseline model, achieving an F1-score of 0.99 for attack detection. We then employ SHAP to scrutinize the operational utility of the generated explanations. Our analysis reveals a critical “last-mile” problem in explainability, demonstrating that raw, unprocessed xAI outputs can be unintelligible to security practitioners due to standard data preprocessing steps like feature scaling. The results show that while xAI can identify influential features, significant postprocessing and contextualization are required to translate these outputs into genuinely transparent and actionable insights, highlighting a crucial gap between theoretical explainability and practical application in cybersecurity.

Memristors have been employed in the design of integrated memory, logic gates, analog circuits, and neuromorphic systems. Their non-volatile nature, high integration density, and compatibility with CMOS technology have been widely documented in numerous studies. Existing research has compared the operational characteristics of various memristor models, including linear, nonlinear, current-controlled threshold adaptive models, and voltage-controlled adaptive models. However, many of these traditional models are either too domain-specific or physically complex, which limits their practical simulation efficiency. In response to the need for a simple, general-purpose, and fast-simulating model in EDA flows, this study proposes an BV-VTEAM which is advanced version of the V-VTEAM model for bipolar memristors. The new model maintains a balance between simplicity and accuracy. In the BV-VTEAM model, the internal state variable represents the memristor's conductivity level. This dynamic variable evolves over time based on the applied stimulus. The state update is performed using a manual integration method, where the time step is automatically selected based on the nearest absolute timestamps in both the growth and decay phases. The bounded function is applied to ensure that internal state variable remains within physical limits. Additionally, the output current is regulated by a maximum threshold using a soft-bounded approach to avoid Zero Jacobian issues, thus improving numerical stability. The operation conditions are analyzed in various dynamic parameters in terms of supply voltage, operating frequency, maximum current, time constant, and nonlinearity exponent This BV-VTEAM model is further utilized to design digital logic circuits, including AND, OR, and flip-flops, laying the groundwork for constructing more complex digital systems based on memristive components.

This paper reviews the research progress on the resilience of coastal cities to typhoon disasters in the context of climate change. It synthesizes advances in theoretical frameworks, quantitative assessment methods, and practical solutions for enhancing resilience to natural disasters. Drawing on complex systems theory, the study conceptualizes disaster resilience and distinguishes it from traditional disaster prevention approaches. The paper examines the strengths and limitations of three primary assessment methods: index systems, scenario analysis, and functional modeling. Furthermore, the study evaluates the adaptability of both structural and non-structural measures to enhance resilience, and it proposes a comprehensive theoretical framework for typhoon resilience in coastal cities. The review also highlights key challenges, including the lack of consensus on theoretical frameworks, unclear mechanisms for assessing resilience to complex natural disasters, and limitations in current multi-hazard assessment approaches. Finally, the paper outlines future research directions, emphasizing the importance of data sharing, the application of big data and artificial intelligence, the development of integrated models, and deeper investigation into resilience mechanisms to support sustainable urban development amid increasing climate uncertainty.

This study presents a concrete stress monitoring method utilizing 1D CNN deep learning of raw electromechanical impedance (EMI) signals measured with a capsule-like smart aggregate (CSA) sensor. Firstly, the CSA-based EMI measurement technique is presented by depicting a prototype of the CSA sensor and a 2 degrees of freedom (2 DOFs) EMI model for the CSA sensor embedded in a concrete cylinder. Secondly, the 1D CNN deep regression model is designed to adapt raw EMI responses from the CSA sensor for estimating concrete stresses. Thirdly, a CSA-embedded cylindrical concrete structure is experimented with to acquire EMI responses under various compressive loading levels. Finally, the feasibility and robustness of the 1D CNN model are evaluated for noise-contaminated EMI data and untrained stress EMI cases.

ABSTRACT In higher education, PLS-SEM supports the study of multidimensional relationships, such as technology acceptance behavior and educational innovation effectiveness. PLS-SEM is a powerful analytical tool used to test complex theoretical models, especially in cases where data are limited or non-normally distributed. TAM provides a theoretical foundation for usefulness and ease of use, while IRT emphasizes psychological and functional barriers to innovation. Data were collected from 420 lecturers at Vietnamese higher education institutions on the acceptance of using artificial intelligence (AI) in higher education to test the integrated model of TAM and IRT. The results show that perceived usefulness (PU) and perceived ease of use (PEU) positively affect attitudes and intentions to adopt AI. In contrast, IRT barriers such as complexity of use, value, risk, and tradition have adverse effects, limiting lecturers’ application of AI in teaching. The research results contribute to the expansion of PLS-SEM applications in higher education and provide recommendations for managers to promote the acceptance of AI technology and reduce barriers to innovation in this field.

Ground reaction in terms of stress and displacement analysis is one of the major challenges in tunnel design and is helpful for understanding the stability around the excavation zone. In this paper, the analysis of stress, strain responses and radial displacement around circular tunnel excavated in cohesive-frictional granular materials is discussed via an innovative multi-scale approach (FEM × DEM) that combined finite element (FEM) and discrete element (DEM) methods and compared to other solutions obtained with conventional constitutive laws. Thanks to simulations, (i) the Ground response curve is then established and provides a better understanding of the complex behaviour in the vicinity of the tunnel; (ii) the prediction by FEM × DEM approach differs considerably from conventional constitutive laws; and (iii) in a unique manner, macroscopic and microscopic analyses are simultaneously performed showing that at the macroscopic level (FE mesh), diffuse failure mode are observed, while at the microscopic level (DEM level), microscopic evolutions are observed to be strictly related to the macroscopic behaviour.